All pull requests

Pull requests

• feat(inference): add utterance_end_ms to DeepgramOptions

  inference-review
  #8498·adrian-cowham opened Nov 10, 2026Contributor·

  ·
  

• feat(avatar): forward the user's speaking state to remote avatar workers

  #8497·ggarber opened Nov 10, 2026Contributor·

  ·

• feat(voice): commit only final transcripts with commit_user_turn(wait_for_final=True)

  #8495·captainbanan opened Nov 10, 2026Contributor·

  ·
  3

• feat(reson8): support rumble-2 STT model

  #8494·phillip-kil opened Nov 10, 2026Contributor·

  ·
  5

• examples: the front desk with a 3rd-party back office

  #8493·longcw opened Nov 10, 2026Contributor·10/10·

  ·
  4

• fix(voice): send a returned tool result when the interruption timeout ends its wait

  #8492·TANGBUDU opened Nov 10, 2026Contributor·

  ·
  1

• feat: add Talking Avatar avatar plugin

  #8491·Tushar1357 opened Nov 10, 2026·

  ·
  4

• feat(inference): Keep a spare connection after a cancelled request

  #8486·shawnfeldman opened Nov 9, 2026Contributor·

  ·
  3

  

• fix(codecs): keep the packets read while probing the stream

  #8484·Kayvan-Zahiri opened Nov 9, 2026·

  ·

• fix(workflows): identify card issuers by IIN range, not the first digit

  #8483·Kayvan-Zahiri opened Nov 9, 2026·

  ·
  2

• fix(cambai): request the TTS sample rate the frames are labeled with

  #8482·GPT-AGI opened Nov 9, 2026·

  ·
  1

• feat(elevenlabs): send similarity and clone_strength to dialogue models

  #8481·ethankraus-elyos opened Nov 9, 2026·

  ·
  1

  5

• fix(room_io): resample pre-connect audio with its channel count

  #8479·GPT-AGI opened Nov 9, 2026·

  ·
  1

• fix(voice): keep the source session usable after a rejected handoff

  #8478·ItsM0rty opened Nov 9, 2026·

  ·
  1

  1

• Port initial realtime session configuration from agents-js

  #8474·rosetta-livekit-bot[bot] opened Nov 9, 2026Bot·

  ·
  3

• fix(nvidia, xai, upliftai): report the provider in metrics

  #8473·anubhavlal07 opened Nov 9, 2026·

  ·
  4

• fix(assemblyai): honor the language passed to STT.stream()

  #8472·anubhavlal07 opened Nov 9, 2026·

  ·
  1

• fix(soniox): honor the language passed to STT.stream()

  #8471·anubhavlal07 opened Nov 9, 2026·

  ·
  1

• fix(assemblyai): tolerate None in language and language_codes options

  #8470·pylrn opened Nov 9, 2026·

  ·
  3

• feat(voice): add AudioOutput.set_volume with a click-free ramp

  #8467·GoGennady opened Nov 9, 2026·

  ·
  1

  4

• fix(elevenlabs): retry a failed stream on a fresh websocket

  #8465·GoGennady opened Nov 9, 2026·

  ·
  1

  3

• fix(responses): emit message phase before text deltas

  #8463·Epsilon006 opened Nov 9, 2026Contributor·

  ·
  1

  7

• tests: assert complete tokenizer output

  #8458·ilakkmanoharan opened Nov 9, 2026·

  ·

• feat(saidwire): add Saidwire STT plugin

  #8457·Hudson6124 opened Nov 8, 2026·

  ·
  12

• feat(zoom-ai): add Zoom Voice Agent realtime model

  #8456·tendant opened Nov 8, 2026·

  ·
  11